ANALYSIS OF HOUSEHOLD VULNERABILITY TO CLIMATE CHANGE AND ADAPTATION
OPTIONS: EVIDENCE FROM ADAMA AND LUME WOREDAS, ETHIOPIA
Bedaso Taye
Sasakawa Global 2000
Monitoring, Evaluation and Learning Program Officer
<bedho250@gmail.com>
Abstract
The study assesses the extent of household vulnerability to climate change by applying Vulnerability as
Expected Poverty (VEP) approach. The VEP approach is based on estimating the probability that a given
shock or set of shocks moves household consumption below poverty line or forces them to stay there if
they are already below poverty line. The data are collected from rural farming households in Adama and
Lume woredas of East Shewa zone during the 2009 production season. The result shows that about 68
percent of farmers are vulnerable to poverty, while 62 per cent of them are observed to be poor. About 52
per cent of the households are vulnerable to poverty due to low consumption prospect and 16 per cent of
them are vulnerable due to high consumption volatility. The study also indicates that change of rainfall and
temperature from long run averages, frequency of drought and soil characteristics are major reasons for
farmers’ vulnerability to poverty. On the other hand, education of the head of the household, livestock and
land size, irrigation size, quantity of fertilizer used and number of extension contacts are found to reduce
household vulnerability to climate change. Proximity to facilities such as road and market also reduces
farmers’ vulnerability. But, higher family size and exposure to non-climatic shocks such as death of a
household head and input price rise increase vulnerability. On top of that, the estimated incidence of
poverty is less than the fraction of population that is vulnerable to poverty. This calls for differential
intervention for poverty reduction and poverty prevention, in addition to consumption stabilization and
increasing measures. On the other hand, expansion of extension services, irrigation practices, non-farm
income opportunities, improvement of farmers’ access to fertilizer use and increase of household capacity
to cope or mitigate shocks and risks are important intervention areas to reduce vulnerability.
Key words: Vulnerability, climate change, vulnerability as Expected Poverty, Adama, Lume
i
Bedaso Taye
1. INTRODUCTION
Agriculture is the main sector of the Ethiopian economy. It contributes 40% of
GDP, generates more than 90% of foreign exchange earnings and employs
about 85% of the population (Kumar and Quisumbing, 2010). However, the
agricultural sector is dominated by small-scale, mixed-crop and livestock
production which is characterized by low productivity. The major factors
responsible for low productivity include reliance on obsolete farming techniques,
soil degradation caused by over-grazing and deforestation, poor complementary
services such as extension, credit, marketing, infrastructure, and climatic factors
like drought and flood (Deressa et al., 2008a). These factors reduce the adaptive
capacity or increase the vulnerability of farmers to future climate change and
variability, which negatively affects the performance of the already weak
agriculture.
A recent mapping on vulnerability and poverty in Africa (ILRI, 2006), puts
Ethiopia as one of the countries most vulnerable to climate change and with
least capacity to respond. Ethiopia already suffers from extremes of climate,
manifested in the form of frequent drought and flood (Difalco et al., 2007). On
the other hand, Ethiopian agriculture is largely rain-fed, with irrigation practices
accounting for negligible portion of the total cultivated land in the country. Thus,
the amount and temporal distribution of rainfall, temperature and other climatic
factors during the growing seasons has an important influence on crop yields
and can induce food shortages and famine which increase farmers’ vulnerability
to poverty.
More importantly, studies in Ethiopia show that frequency of droughts and its
spatial coverage have increased over the past few decades (Deressa et al.,
2008a). According to World Bank (2008), Ethiopia has experienced at least five
major national droughts since 1980 along with literally dozens of local droughts.
Cycles of drought create poverty traps for many households, constantly
thwarting efforts to build up assets and increase income. World Bank survey
data show that between 1999 and 2004 more than half of all households in the
country experienced at least one major drought shock. These shocks are a
major cause of transient poverty. Had households been able to smooth
consumption, then poverty in 2004 would have been at least 14% lowera
figure that translates into 11 million fewer people below the poverty line (World
Bank, 2008).
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
On top of that, the average annual minimum temperature over the country has
been increasing by about 0.250C every ten years, while average annual
maximum temperature has been increasing by 0.1 0C every decade. There is
also a decreasing precipitation over the country (NMSA, 2001). The past trends
of increasing temperature, decreasing precipitation and increasing frequency of
droughts are predicted to sustain in the future in the tropics, of which Ethiopia is
one (IPCC, 2001 in Deressa et al., 2008a). Therefore, the country’s agriculture is
exposed to adverse climate conditions and, thus, vulnerable to climate change.
Estimating vulnerability to climate change and tracking its correlates is important,
because a capable social policy should go beyond poverty alleviation in the
present and examine poverty prevention in the future. A poverty reduction
strategy that ignores the transient nature of poverty misses households that
have a high probability of being poor and may instead devote scarce resources
to households that are only transiently poor and could have found a way out of
poverty without government assistance (Shewmake, 2008). Therefore,
investigating the extent of vulnerability to poverty and understanding its
correlates is important to formulate thriving social policy.
Estimation of vulnerability at the household level should ideally be attempted
with panel data of sufficient length and richness (Chaudhuri et al., 2002).
However, such data are rare, particularly in poor developing economies like
Ethiopia. Instead, the best one can usually hope for are cross-sectional
household surveys with detailed data on household characteristics, consumption
expenditures and income. Cross-sectional data are useful to measure only
variation in welfare at a given point in time, but they are nonetheless an
important analytical tool used to identify risks and vulnerable groups, assess the
outcome and impact of shocks, and identify households that face high risk of
falling into poverty due to climate change. This is especially true if variation in
welfare across households is mainly attributed to observable household
characteristics.
There are several studies conducted to investigate the vulnerability of Ethiopian
farmers to poverty and climate change (Dercon, et al., 2005; Skoufias and
Quisumbing, 2003; Dercon and Krishnan, 2000; Di Falco et al., 2008 and
Deressa et al., 2008a) and suggested policy options to reduce vulnerability.
Many of these studies analyzed vulnerability of farmers to climatic extremes and
non-climatic shocks. But this study will estimate vulnerability of farmers to
3
climate change at household level and examine factors that are responsible for
their vulnerability, including change in climatic elements such as average
temperature and precipitation. Moreover, it takes adaptation by farmers as an
explanatory variable. Therefore, this study aims to assess household
vulnerability to climate change and outline and explain factors that account for
their vulnerability.
The rest of this paper is organized as follows. Section two discusses theoretical
and empirical literatures. The third section presents conceptual framework and
methodology of the study. Section four presents results and findings of the study
and the last section concludes and forwards policy implications of the study.
2. REVIEW OF LITERATURE
2.1. Measuring Vulnerability to Climate Change
There are various approaches to measure vulnerability depending on the
purpose, field and threshold used for assessment of vulnerability. For example,
according to IPCC, Vulnerability to climate change is the degree to which
geophysical, biological and socio-economic systems are susceptible to and
unable to cope with adverse impacts of climate change (IPCC, 2001). Based on
IPCC’s definition, vulnerability is measured as the probability of falling below
some specified threshold, usually poverty line. The most common method
employed in vulnerability assessment is the Econometric Method. The
econometric method has its roots in the poverty and development literature. This
method uses household-level socioeconomic survey data to analyze the level of
vulnerability of different social groups (Deressa et al., 2008a). The method is
divided into three categories: (a) Vulnerability as Expected Poverty (VEP), (b)
Vulnerability as Low Expected Utility (VEU) and (c) Vulnerability as Uninsured
Exposure to Risk (VER) (Hoddinott and Quisumbing, 2003). All three share
common characteristics, in that they construct a measure of welfare loss
attributed to shocks.
2.1.1. Vulnerability as Expected Poverty (VEP)
In the expected poverty framework, vulnerability of a person is conceived as the
prospect of that person becoming poor in the future, if currently not poor or the
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
prospect of that person continuing to be poor if currently poor (Deressa et al.,
2008a). Thus, vulnerability is seen as expected poverty, and consumption
(income) is used as a proxy for well-being. This method is based on estimating
the probability that a given shock, or set of shocks, moves consumption by
households below a given minimum level (e.g., consumption poverty line) or
forces the consumption level to stay below the given minimum requirement, if it
is already below that level (Chaudhuri et al., 2002)1.
The merit of this vulnerability measure is that it can be estimated with a single
cross section data. However, the measure correctly reflects a household’s
vulnerability, only if the distribution of consumption across households, given the
household characteristics at one time, represents the time-series variation of
consumption of the household (Gaihai and Imai, 2008). Hence, this measure
requires a large sample in which some households experience a good period
and others suffer from negative shocks. Also, the measure is unlikely to reflect
unexpected large negative shocks if we use the cross section data for a normal
year. Moreover, if estimations are made using a single cross section, one must
make a strong assumption that cross-sectional variability captures temporal
variability (Hoddinott and Quisumbing, 2003).
2.1.2. Vulnerability as a Low Expected Utility (VEU)
Ligon and Schechter (2003) defined vulnerability as the difference between the
utility derived from some level of certainty-equivalent consumption,
at and
above which the household is not considered vulnerable, and the expected utility
of consumption. This certainty-equivalent consumption is analogous to a poverty
line. Consumption of a household
has a distribution in different states of the
world So this measure takes the form:
=
(
Where,
)-
(
),
is a (weakly) concave, strictly increasing function.
However, measuring vulnerability as a low expected utility requires specification
of a particular utility function which will affect the magnitudes calculated.
Moreover, Hoddinott and Quisumbing (2003) added, while the magnitudes are
1
This method is discussed in detail in sections 3.2 and 3.4 below.
5
affected by changes in functional form, it appears that the relative magnitudes of
the individual components are not so affected. A more problematic concern here
is that the units of measurement are units of utility (e.g., utils); for example a
finding that Vh =0.25 means that the utility of a household, h, is 25 per cent less
than would be the case, if all inequality of consumption and risks in consumption
were eliminated. For many policymakers, this expression of magnitude may be
difficult to understand. Deressa et al. (2008a) added that in this method it is
difficult to account for an individual’s risk preference, given that individuals are ill
informed about their preferences, especially those related to uncertain events.
2.1.3. Vulnerability as Uninsured Exposure to Risk (VER)
The VER method is based on ex-post facto assessment of the extent to which a
negative shock causes welfare loss (Hoddinott and Quisumbing 2003). In this
method, the impact of shocks is assessed by using panel data to quantify the
change in induced consumption. In the absence of risk management tools,
shocks impose a welfare loss that materializes through reduction in
consumption. The amount of loss incurred due to shocks equals the amount paid
as insurance to keep a household as well off as before any shock occurs. The
disadvantage of this method is that, in the absence of panel data sets, estimates
of impacts, especially from cross-sectional data, are often biased and thus
inconclusive (Deressa et al., 2008a). Therefore, it is difficult to apply this
approach with single round cross section data.
2.2. Empirical Literature
Studies on impact of climate change on agriculture at household level are very
scanty. Most of them focused on the effect of climatic extremes on farmers and
the determinants of farmers’ adaptation techniques. For instance, Yesuf et al.
(2008) conducted an empirical analysis of the impact of climate change and
adaptation on food production in the Nile Basin of Ethiopia. They have used
cross section data from 1000 farms producing cereal crops in the basin and
monthly rainfall and temperature data that were interpolated to get household
specific values. They estimated production function into which adaptation
entered as a binary variable. Finally, they concluded that climate change and
climate change adaptation have a significant impact on farm productivity.
Extension services, both formal and farmer to farmer, as well as access to credit
and information on future climate change affect adaptation decision positively
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
and significantly. They also found that farm households with larger access to
social capital are more likely to adopt yield-related adaptation strategies. From
this study, we can get information on the important role of adaptation to climate
change in stabilizing farm productivity and factors that dictate the use of
adaptation options. But we get little information on how climate change affects
household welfare in the coming periods and the role of adaptation in reducing
household vulnerability to future climate change.
Deressa et al. (2008b) assessed the vulnerability of Ethiopian farmers to climate
change, based on integrated vulnerability assessment approach using
vulnerability indicators constructed by principal component analysis. The
vulnerability indictors consist of different biophysical and socioeconomic
attributes of seven agricultural-based regional states. They found that the
relatively least developed arid and semi-arid regions of Afar and Somali are the
most vulnerable to climate change. Tigray and Oromia regions are also
vulnerable to climate change. Therefore, investing in irrigation in the relatively
least developed regions of Afar and Somali, coupled with early warning systems
and production of drought tolerant varieties of crops and livestock, can all reduce
vulnerability of Ethiopian farmers to climate change. In construction of
vulnerability indicator, the different socio economic and biophysical indicators of
vulnerability of each region are classified according to IPCC’s definition of
vulnerability, which consists of adaptive capacity, sensitivity and exposure. But,
arbitrary and subjective weights are attached to different indicators, which
threaten the reliability of the indices. Moreover, the scale of analysis was
regional, which makes the indices too crude to launch policies that are useful to
reduce vulnerability at local or household level.
A study by Sharon Shewmake (2008) uses farmers’ responses to exogenous
weather shocks in South Africa’s Limpopo River Basin to gauge how farmers opt
to respond to future climate change-induced shocks, in particular drought.
Droughts are expected to increase in both frequency and intensity as a result of
climate change. This study examined the costs of drought today, and whom it
affects the most, in an effort to guide policy options in the future. The study used
a combination of descriptive statistics and econometric analysis to approximate
the potential impact of droughts on rural South African households. The study
also estimated household vulnerability to climate change. After controlling for
household heterogeneity using propensity score matching, Sharon Shewmake
(2008) noted that there is no statistically significant impact of droughts on
7
income, thus suggesting households have already adapted to living in a droughtprone environment. The types of households that were more vulnerable to
climate shocks are analyzed, using two measures of vulnerability: the probability
of falling below income of 7,800 South African Rand (R), and the probability of
income falling below 16,000R. Residents of the Limpopo province were the least
vulnerable under both metrics. Setswana and SeSwati households were more
vulnerable than other ethnic groups. Households that do not own livestock and
households that rely on rain-fed agriculture were also more vulnerable than other
households. In this study, climate change is proxied by occurrence of drought,
which hardly captures all components of climate change. Change in
temperature, rainfall, wind and relative humidity are all components of climate
change; ignoring them will lead to biased conclusions about impact of climate
change on agricultural households.
Chaudhuri et al. (2002) noted the importance of cross-sectional data in
estimating household vulnerability to poverty and gave detailed methodological
description and estimates from Indonesia. They stressed that, despite the
obvious limitations of purely cross-sectional data, a detailed analysis of these
data can potentially be informative about the future. If most of the observed
cross-sectional variations in consumption levels across households stem from
unobserved (to us) differences across households, say because of unobserved
household-specific determinants of consumption levels that are persistent over
time, then, clearly, we would not be able to assess household vulnerability to
poverty with any degree of confidence. If, on the other hand, much of the
variation can be attributed to the differences in the observable characteristics of
households, then even a single cross section can be quite helpful in answering
questions about household vulnerability.
In their paper, (Chaudhuri et al., 2002), starting with a definition of vulnerability at
household level, as the probability that a household, regardless of whether it is
poor today, will be consumption poor tomorrow, they provided a conceptual
framework for thinking about the different dimensions of vulnerability to poverty,
and then proposed a simple method for empirically estimating household-level
vulnerability, using cross-sectional data. They also demonstrated the uses and
limitations of the proposed methods through a case study using household-level
data from Indonesia.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
From the cross-sectional study in Indonesia, they drew three main conclusions.
First, the fraction of the population that faces a non-negligible risk of poverty is
considerably greater than the fraction that is observed to be poor. While 22% of
the Indonesian population was observed to be poor, they estimated that 45% of
the population was vulnerable to poverty. Second, the distribution of vulnerability
across different segments of the population can differ markedly from the
distribution of poverty. They argued that this highlights the need for a distinction
between programmes of poverty preventionthose aimed at reducing
vulnerabilityand poverty alleviation, and for differential targeting of the two.
Third, they found striking differences in the sources of vulnerability for different
segments of the population. For rural households and for less-educated
households, the main source of vulnerability appears to be low mean
consumption prospects; for urban households and for more highly educated
households, on the other hand, vulnerability to poverty stems primarily from
consumption volatility.
3. CONCEPTUAL FRAMEWORK AND METHODOLOGY OF THE STUDY
3.1. Conceptual Framework
The conceptual framework for this study depends on the IPPC’s (2001) definition
of vulnerability to climate change. The IPCC defines vulnerability to climate
change as follows: “The degree to which a system is susceptible, or unable to
cope with adverse effects of climate change, including climate variability and
extremes, and vulnerability is a function of the character, magnitude and rate of
climate variation to which a system is exposed, its sensitivity, and its adaptive
capacity”. See Figure 1 below.
As Figure 1 shows, farmers are exposed to both gradual climate change (mainly
changes in temperature and precipitation) and extreme climate conditions
(mainly drought and flood). Exposure affects sensitivity, which means that
exposure to higher frequencies and intensities of climatic risk highly affects
outcome (e.g., yield, income, health). Exposure is also linked to adaptive
capacity. For instance, higher adaptive capacity reduces the potential damage
from higher exposure. Sensitivity and adaptive capacity are also linked: Given a
fixed level of exposure, the adaptive capacity influences the level of sensitivity.
Adaptive capacity reduces socioeconomic vulnerability, vulnerability that results
9
from the socioeconomic and political status of an individual or household.
Individuals in a community often vary in terms of education, gender, wealth,
health status, access to credit, access to information, technology, social,
environmental and physical capital; political power, and so on. These variations
are responsible for the variations in vulnerability levels. On the other hand,
sensitivity of a system to environmental stresses increases its vulnerability,
biophysical vulnerability, which is the level of damage that a given environmental
stress (factor) causes on both social and biological systems (example, impact of
climatic variables and soil characters on vulnerability). Therefore, vulnerability to
climate change is affected by the sensitivity of the system to the change and its
adaptive capacity that are, in turn, a function of other biophysical, climatic and
household characteristics.
Climate change
(Gradual)
o
o
o
o
o
o
Increase in
temperature
Fall in rainfall
Climatic extremes
o
o
o
Exposure
Drought
Flood
Hailstorm
etc
Sensitivity
Adaptive capacity
Biophysical
vulnerability
Socioeconomic
vulnerability
o
Soil
Rainfall
Temperature
Water availability
o
o
o
Resource
endowments
Adaptation
Household size
Education etc.
Total vulnerability
Figure 1. Conceptual Framework of Vulnerability Assessment
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
3.2. Empirical Model and Sources of Data
3.2.1. Empirical Model
Among the above discussed approaches to measure vulnerability, the probability
of a household to fall below a given consumption level due to climate change will
be measured by adopting vulnerability as expected poverty approach (VEP).
This will help us to know the proportion of farmers that are vulnerable to climate
change and hence attribute vulnerability to different factors, including adaptation
decision to set up policy options to reduce vulnerability.
Following Chaudhuri et al. (2002), the stochastic process generating the
consumption of a household h is given by:
(1)
Where, Ch is per capita consumption expenditure,
represents a vector of
observable household characteristics such as household size, location,
educational attainment of the household head, land size, non-farm income etc.,
climatic factors and shocks such us temperature, precipitation, drought, flood
and adaptation strategies, β is a vector of parameters to be estimated and
a mean zero disturbance term.
is
The probability that a household will find itself poor depends not only on its
expected (mean) consumption but also on the volatility (i.e. variance, from an
inter-temporal perspective) of its consumption stream (Jamal, 2009). Therefore,
household expected consumption and the variance of its consumption are
required to quantify the level of households’ vulnerability to climate change.
Cross section consumption variance is estimated from the error term as follows.
Assume that the variance of
is given by:
(2)
β and θ are parameter estimates from a three-step Feasible Generalized Least
Squares (FGLS) procedure suggested by Amemiya (1977). First, Equation (1) is
estimated using an Ordinary Least Square (OLS) procedure. The residuals
from equation (1) are then regressed on
11
using OLS as follows:
=
+
(3)
The predicted values Xh
transform Equation (3.6).
from this auxiliary regression are then used to
=
(4)
This transformed equation is estimated using OLS to obtain an asymptotically
efficient FGLS estimate (
estimate of
). It can be shown that (
) is a consistent
, which is the variance of the idiosyncratic component of the
household consumption. Equation (3.4) is also transformed with the standard
error of (
).
=
(4)
=
(5)
OLS estimation of Equation (5) yields a consistent and asymptotically efficient
estimate of
. The estimated
and
symbolize the expected log
consumption and variance of log consumption respectively. The expected log of
consumption and variance of log consumption for each household h are,
respectively, estimated as:
=
=
(6)
2
e,h=Xh
(7)
By assuming that consumption is log normally distributed (i.e.
is normally
distributed), the above enable estimation of the probability that a household with
the characteristics
will be poor, i.e. a household’s vulnerability level. Letting
denote the cumulative density of the standard normal, the estimated
probability will be given by:
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
h=
=
(8)
Where,
is the log of the minimum consumption (income level) beyond which
a household would be vulnerable.
The above analysis is based on the assumption that experiencing different
climatic conditions and shocks such us drought, floods and hailstorm will
increase the probability of farmers falling below a given consumption or income
level or force them to stay under the poverty line, if they are already there.
From (8) we can get the level of vulnerability of the household to poverty, and
hence we can classify households according to their level of vulnerability. So,
while vulnerability is a risk and comes in degrees (between zero and one), being
vulnerable is a state (either zero or one). Using the argument forwarded by
Pritchett et al. (1999), this study will take the threshold probability level that
defines a vulnerable household to be 0.5. This has two attractive features. First,
50-50 odds is a nice “focal” point and it makes intuitive sense to say a household
is “vulnerable” if it faces more than 0.5 probability to be poor. Second, if a
household is just at the poverty line and faces a mean zero shock, then this
household has a one period ahead vulnerability of 0.5. This implies that, in the
limit, as the time horizon n goes to zero, then being “in current poverty” and
being “currently vulnerable” coincide.
Given that vulnerability of households is a bounded variable between 0 and 1, in
order to use the OLS regression it needs to be transformed to a positive
unbounded variable. Following literature one transformation is to calculate the
variable U, where,
The problem with this transformation is related to the fact that U is not normally
distributed, since most of the values are concentrated between 0.9 and 1. In
order to smooth this problem, the natural logarithm of U is used instead. The
final formula for the dependent variable, thus, becomes;
13
Once a dependent variable is defined like this, the explanatory variables are in
levels, except the estimated value of assets other than land which is in log form.
Therefore, the interpretation of marginal effect of the independent variables is in
percentage form since the model is semi log.
In this study, international poverty line of USD 1 is used. Normally, national
poverty line is important because it reflects local conditions and serves as a
basis of development planning and policy making. However, the recent national
poverty line of ETB 1075 per year per adult equivalent that was established in
2004/05 by MoFED cannot reflect the current market price of goods and
services. Moreover, the study intends to estimate the impact of climate change
on households’ vulnerability, thus the use of USD 1 as a threshold is not serious
limitation. Accordingly, international poverty line of USD 1 is equivalent to ETB
4330.56 per year per adult equivalent using average exchange rate of 2009
(survey year).
3.2.2. Data Sources and Sampling Techniques
The data used in this study is obtained from a household survey of production
period 2009 in rural kebeles2 of Adama and Lume woredas in East Shewa zone.
The sample kebeles are purposely selected to include different attributes in the
area. The attributes include average rainfall, rainfall variability, food aid
dependent population due to droughts and rainfall fluctuation, irrigation activity,
availability of meteorological data etc. The survey covers 10 kebeles from two
woredas. By considering the socioeconomic and environmental conditions, 8
kebeles from Adama woreda and 2 kebeles from Lume woreda were selected for
this study.
Once the kebeles are identified, households are selected based on the total
households in the kebele. Sample households are selected using systematic
sampling method by picking every Nth household, starting from a random start.
The survey covers a total of 230 farming households, out of which 222 are free
of errors and omissions, thus used in the analysis.
2
Kebele is the smallest administrative unit in the federal structure. It comprises different
gots/ketenas. Woreda is the third administrative level in the federal tiers (federal, region, woreda,
kebele). It comprises different kebeles.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
Secondary information and data used in the analysis are mean monthly
maximum and minimum temperature and mean monthly rainfall values and
types of soils. Data on rainfall and temperature were obtained from Adama
meteorological branch office that has 139 meteorological stations under it.
Average monthly minimum and maximum temperature and rainfall data were
available since 1978 from many of these stations and since 1988 from some of
them. Mean monthly maximum and minimum temperature and monthly rainfall
data of 5 stations that coincides with the above sample kebeles, were obtained.
These are Adama, Sodere, Koka dam, Welenchiti and Modjo stations.
4. RESULTS AND DISCUSSIONS
4.1. Descriptive Statistics
The survey covers a total of 222 households that are engaged in crop production
and rearing of livestock as their primary activity. Female-headed households
comprise 15 per cent (33) of the total sample. The average family size in the
sample is 5.91 persons per household, which is comparable with the regional
level figure of 5 persons per household and particularly closer to East Shewa
zone’s figures of 5.13 persons per household. Adult equivalent 3 scale for
average family size is 5.09. Dependency ratio is 0.85 for the households,
showing presence of small number of dependent members in the households.
Most of the dependents are children less than 14 years of age. Average
household head age in the sample is 43.88 years and an average year of
schooling completed by household heads is 3.56 years.
The sample households mostly keep animals like oxen, cows, sheep, goat,
donkey, poultry etc. Average livestock holding per head is 6.66 TLU 4. Ox is a
more important livestock held by many for draught power, followed by donkey for
transportation. Most households also keep poultry, sheep and goats which they
bring to the market, especially during crop failure. The survey also included
enumeration of the estimated value (in ETB) of other assets possessed by
households other than land and livestock. The estimated value of assets owned
3Adult Equivalent is estimated as per scale given by Krishnan and Dercon (1985).
4 TLU is computed using the following conversion factors, cow and ox = 1, heifer = 0.75, calf =
0.25, donkey = mule = 0.70, horse = 1, camel = 1.2, goat and sheep = 0.13, poultry = 0.013 (Stork
et al., 1992).
15
by households in the sample is ETB 19,351.68 per household. The other asset
that household possess and is perhaps most important in the area is cultivable
land on which they cultivate crops. Per capita landholding in the sample is 1.83
hectare per household. There is considerable variation in land size among
households ranging from 0 to 8 hectares per household. There are also
households that have access to irrigable land and the average holding per
household is, however, very small than the non irrigable land. About 12 per cent
of the households have access to irrigation water which lies along the Awash
River. Irrigated landholding is 0.11 hectare per household. In the survey area,
crop production is conducted only during the main rainy season that runs from
May to September/November. Apart from those who have access to irrigation, all
the sampled households produce once in a year.
The major soil type in the study area is sandy loam, which covers 50 per cent of
the woreda, and more than 70 per cent of the study area. Vertisols and Andosols
also cover a significant portion of the land, 7 and 23 per cent, respectively, of the
study area. All these types of soil share a common characteristics in that they
have low moisture retaining capacity, but highly productive when there is
adequate moisture. The real threat to this is, however, the increasing soil
degradation and erosion due to the looseness of the soils that allows them to be
easily taken away by erosion forces.
Differential access to resources and services is another aspect of the study area.
Access to social utilities and infrastructure is relatively better in the study area.
On average, all households have to travel 8.34 km (two ways) on average to get
an all-weather road and 12.29 km to get services of secondary schooling,
banking, hospitalization, daily input and output market. Extension services are
one of the most easily available technical supports to farmers these days in
Ethiopian context. All villages have access to extension services provided by the
nearest Development Agent (DA) Office. Farmers travel less than a kilometre to
get extension services. Variation in use of extension services is, however,
observed across households. On average all households have made 7.55
extension contacts during the 2009 (2001/02 E.C) production season. The
maximum contact is observed to be 52 times per season and the minimum is 0.
In terms of input use, farmers in the study area used 349.62kg of fertilizer (DAP
and urea) and 1.61 litres of herbicide and pesticide per head.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
Access to credit is another important aspect in rural livelihood improvement
because it is an intermediate income that increases productivity and generates
income. According to the survey, during the 2009 production season, 28.83 per
cent of the farmers had access to credit facility. Non-farm income is also
important for rural livelihood both during crop failure and good harvest. In our
case, 35 per cent of the households have one or more non-farm income, which
amounts to at least ETB 1000 per year. The sources of their income are
remittances, sale of trees, charcoal, self employment and petty trade among
others. Most non-farm income earners derive their income from sale of charcoal
(25%), land rental (14.6%), petty trade (11%), self employment (10.4%), aid or
donation (7.9%).
The survey also covers whether a household suffers non climatic shocks or not.
Accordingly the most frequently reported shocks are input price rise beyond their
expectation (68%), pest and disease (33%), output price fall (32%) and animal
disease (31%). According to the survey, these shocks have resulted in loss of
household welfare. For instance, 65 per cent of households that reported input
price rise have claimed that it has affected the household ‘very negatively’, and
30 per cent affected ‘negatively’ while 5 per cent of them are ‘not at all affected’
by the shock.
4.2. Climate of the Study Area
Rainfall
The study area is mostly kolla5 (70%) and the rest (30%) is Woina Dega. The
rainfall pattern of the study area shows high variability. Variance of long run
average annual rainfall is 365, indicating high variability from year to year. Figure
2 shows deviation of mean annual rainfall from the long run average. The long
run average rainfall computed for the period 1978 to 2009 is 72.93mm. As it can
be seen from figure 2, annual mean rainfall fluctuates around this value.
In the study area, long run average maximum and minimum temperatures are
28.86 and 13.59 degree Celsius, respectively. The variability of maximum and
minimum temperatures is 0.935 and 0.726 degree Celsius, respectively. The
5
Note: definition of agro-ecologies is as follows: Dega: Altitude: 2500-3000masl and Rain fall:
1200-2200mm, Woina Dega: Altitude: 1500-2500 masl and Rain fall: 800-1200mm, Kolla:
Altitude: <1500masl and Rain fall: 200-800mm.
17
deviation of maximum and minimum temperature from the long run average is
not as high as that of rainfall, but there are fluctuations around long run averages
that affect crop production.
Figure 2. Mean Annual Rainfall Values for Selected Stations in
Adama and Lume Woreda
It will be worthwhile to add here the pattern of relationship between total crop
production and the mean annual rainfall in the study area. Figure 3 gives this
relationship for the last decade in Adama woreda. As hypothesized in this study,
the relationship between rainfall and annual crop production in the area follow
each other. When there is a fall in rainfall, total crop production also falls. This
could be a possible place to start analysis of relationship between household
welfare and rainfall.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
Figure 3. Relationships between Total Crop Produced and Rainfall for the Last
Decade in Adama Woreda (1999-2009)
The survey instrument also tracked farmers’ perception and understanding on
climate change, their adaptation and coping mechanisms. They were asked
whether they have observed changes in mean temperature, rainfall, frequency of
droughts; flood etc. over the last 20 years. About 94%, 95% and 75% per cent of
them have indeed observed change in temperature, rainfall and frequency of
droughts, respectively, over the last 20 years. Moreover, about 87%, 89% and
75% of them reported respectively that temperature has increased, rainfall has
decreased and frequency of drought has increased. These indicate that farmers’
perception is consistent with the meteorological stations data. The farmers also
reported that these changes have affected their welfare negatively, through crop
loss, livestock loss, health loss and other assets loss.
The most widely practiced adaptation techniques among the households is soil
conservation, adopted by 45 per cent of households, followed by planting of
trees (35%) and adoption of improved farming techniques that include early or
late planting and improved seed application. We should, however, note here that
19
farmers practice soil conservation or planting of trees not only for ameliorating
impact of climate change, but because of government or NGO initiatives and
profit motive. When faced with bad harvest due to rainfall shortage their main
coping mechanism is selling of their livestock and out-migrating in search of offfarm income earning opportunities.
Coping mechanisms are more common during shortage of rainfall and drought.
They are direct responses to climate change than adaptation. Selling livestock is
the major coping mechanism among the households (35%), participation in food
for work and searching of non-farm income are also widely practiced.
4.3. Consumption Expenditure and Estimates of Vulnerability
The survey indicates that the average consumption expenditure of a household
is ETB 19,843 per year. This equals ETB 2981 per year per adult equivalent,
which means $0.6 per day per adult equivalent. Food and non-food expenditure
accounts for about 74 and 26 per cent, respectively, of the total expenditure.
Increased (higher) proportion of non-food expenditure indicates rising living
standard of households. When disaggregated for kolla and woina dega agroecologies, consumption expenditure of households in kolla areas has a higher
average expenditure (ETB 20,705) relative to farmers in woina dega areas (ETB
17,847). But consumption variance in kolla areas is higher than the consumption
variance in woina dega areas, indicating less variability of consumption in woina
dega than in kolla areas. However, the figures provided here should be
interpreted taking into account the possible bias in consumption measurement
especially in cases where consumption data is based on a single-visit interview.
On the other hand, the share of non-food expenditure in total expenditure is
higher in kolla (35%) than in woina dega agro-ecology (32%).
4.3.1. Incidence of Vulnerability and Poverty
The estimated head count vulnerability among the households shows 68.02 per
cent of them have 50 per cent or more probability to become poor next year,
using USD 1 per day per adult equivalent as poverty line 6. On the other hand,
the incidence of poverty among the households is 62.16 per cent, using the
6
One dollar poverty line is converted to ETB using the daily average of 2009 inter-bank exchange rate of USD
to ETB.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
same benchmark of USD 1. This shows that there are households that are now
regarded as being non poor, but with high probability of becoming poor next
year. These estimates, however, are highly sensitive to poverty line, for instance
if poverty line is fixed at USD 0.6, households that are vulnerable to poverty
become only 4 per cent while those regarded as poor become 14 per cent.
Vulnerability of farmers varies between kolla and woina dega agro-ecologies. In
kola, the extent of vulnerability and poverty is 63.87 and 58.71 per cent
respectively, whereas in woina dega areas these figures are 77.61 and 70.15
per cent respectively. Table 1 provides estimates of vulnerability to poverty for
households. The vulnerability rate of 68.02 per cent is more than the estimated
head count poverty level of 62.16 per cent. This indicates that the estimated
probability of experiencing poverty in the near future is more than the observed
incidence of poverty in the sample. Thus the observed incidence of poverty
underestimates the fraction of population that is vulnerable to poverty. The level
of underestimation is revealed by the vulnerability to poverty ratio which is 1.09
for the total sample.
Table 1. Estimates of Vulnerability to Poverty
Estimates of Vulnerability to Poverty
[households with vulnerability >0.5]
Agroecology
Percentage of population
vulnerable
poor
Kolla
63.87
W. Dega
77.61
58.71
70.15
62.16
Total
68.02
Source: Estimated from survey data.
vulnerability
poverty ratio
1.09
1.11
1.09
Relative vulnerability is also used as another measure of incidence of
vulnerability among households. Relative vulnerability uses incidence of poverty
among the households as the threshold, above which household is regarded as
vulnerable. Accordingly, 29.28 per cent of the households have more than 62.16
per cent (head count poverty incidence) probability to become poor next year. As
Table 2 shows, relative vulnerability is high in woina dega than in kolla areas.
Table 2 gives a cross distribution of the percentage of vulnerable and poor
households. It is evident from the table that a significant percentage of the non-
21
poor will become vulnerable to poverty next time. About 42 per cent of the nonpoor households are estimated as being vulnerable to poverty. Obviously, a
majority of the poor (84%) are also vulnerable to poverty. This suggests that
programmes that aim to reduce vulnerability should be designed and targeted
differently from those aimed at poverty alleviation.
Table 2. Cross-Distribution of Vulnerability and Poverty
Cross-Distribution of Vulnerability and Poverty
Poverty status
Vulnerability status
Vulnerable
Nonvulnerable
Poor
Non poor
Total
Source: Estimated from survey data.
84.06
41.67
68.02
Total
15.94
62.16
58.33
37.84
31.98
100.00
According to Chauduri et al. (2002), the sources of vulnerability of households
are low mean consumption and high consumption volatility. To identify source of
vulnerability of households, we classified the households into three groups. The
first group are those households that have vulnerability estimate below 0.50 and
consumption above the poverty line. These are households that are nonvulnerable and non-poor. The second group are households that are non-poor
but vulnerable. These households are vulnerable because of high consumption
volatility; were we able to eliminate the variability in their consumptions, these
households would be no longer vulnerable. The third group are households that
are vulnerable as well as poor. These are households that are vulnerable due to
low consumption prospect. These households have vulnerability level above 0.5,
and their vulnerability stems primarily from their low levels of mean consumption,
in that reduction in consumption volatility would still leave them vulnerable.
We estimated that households that are vulnerable due to high consumption
volatility are only 16 per cent, whereas 52 per cent are due to low mean
consumption. So, vulnerability is mainly due to low consumption prospects and
thus reduction in vulnerability should start from increasing household
consumption.
4.3.2. Econometric Results
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
An econometric model is estimated to show the impact of various household and
environmental factors on household vulnerability to poverty. Assuming linear
relationship between household and environmental factors, there are 31
variables that are included in the model. Model information shows the model is
suitable for the problem at hand. Adjusted R squared and over all F value are
0.71 and 18.61 respectively. Before inclusion of variables into the econometric
framework, the variables are refined by ANOVA to see difference in mean values
of variables between vulnerable and non-vulnerable households. Table 3 gives
model information and the OLS estimates of the coefficients of the model.
Household consumption is often modelled as a linear function of household
characteristics. By assuming that vulnerability is also linearly related with
household and environmental characteristics, the estimated vulnerability function
is as presented below. Most of the variables are significant with expected signs.
To avoid the possible hetroskedasticity problem, Feasible Generalized Least
Square (FGLS) estimates are provided. The model is free of multicollinearity
problem as mean VIF is 2.00.
Family Composition
There are two controversial views on the relationship between family size and
the welfare of the household. The first argument states that households who
have larger family size are supposed to be better off than those having smaller
family size, since there are advantages in consumption economies of scale and
availability of more working labour force to generate income (Adane and
Bezabih, 2003). In contrast to this, there is another convincing argument that a
family size increases the probability that a household falls below poverty line due
to having more people leading to disguised unemployment due to scarcity of
capital and also due to increased dependency ratio. The finding of the study has
supported the second idea, in that as the family size increases vulnerability of
households to poverty also increases. As family size increases by 1 on average,
household vulnerability to poverty increases by about 5.83 per cent and the
estimate is significant at 1 per cent significance level.
The age structure of the household head is another area that should get due
consideration since it has an important implication on economic productivity,
experience and asset endowment. The mean age of a household head age is
45.47 for the vulnerable group and 40.51 for the non-vulnerable group and the
23
difference is statistically significant. Probably as a farmer gets older, he can
acquire farming experience throughout his life that could have a positive
contribution in raising his living standard. Similar pattern emerges from this
study; as a household head’s age increases by one year, household vulnerability
to poverty decreases by 1.48 per cent approximately.
Household Education
A number of studies have indicated that the household head is a highly
influential decision maker in the Ethiopian family. This suggests that his/her
education level does matter for the welfare of the family. Astonishingly, the result
of the study indicates that the education level attained by the household head is
low. Tangibly, average numbers of years of school attended by the vulnerable
group household head is 3.17, whereas it is 4.38 for the non vulnerable and the
difference is statistically significant at 5% level. Household education is thus a
significant factor that explains difference in household vulnerability levels. The
econometric result also indicates that a household head’s and spouse’s
education reduces household vulnerability to poverty. As the head’s years of
education and spouse’s years of education increases by 1 year, household
vulnerability to poverty decreases by 3.16 and 14.86 per cent respectively.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
Table 3. Correlates of Estimated Vulnerability Level of Households
Estimated vulnerability Function- FGLS Estimates
[Dependent variable- vulnerability level of households: Equation 8]
Variables
Intercept (Constant)
Household Demography
Family Size
Dependency Ratio
Age of Head
Sex of Head
Household Education
Education of Head
Education of Spouse
Household Assets
Log of value of other assets
Livestock
Land Size
Irrigation size
Inputs and Facilities
Road
Market dummy
Extension Contacts
Credit
Non Farm Income
Idder
Equb (traditional saving group)
Fertilizer
Climate, Adaptation and Soil
Rain Fall Change
Minimum Temperature
Soil dummy
Drought
Coefficients
-0.05006
t-statistics
-0.19
0.05832
-0.03966
-0.01478
0.03087
4.00***
0.93
-5.14***
0.41
-0.03163
-0.14866
-3.33***
-16.09***
-0.02811
-0.04008
-0.20282
-0.27656
-1.44
-10.00**
-9.03***
-3.87***
-0.01775
-0.01465
-0.01518
0.00219
-0.02674
-0.34926
-0.52342
-4.71***
-3.27***
-4.99***
1.28
-7.62***
-3.03***
-7.79***
-0.00067
-4.89***
-0.25003
0.47201
0.25477
0.13632
-4.56***
8.60***
2.18**
8.94***
Adaptation
-0.09955
-1.96**
Shocks
Input price rise
0.09475
1.66*
Death of Household Head
0.2784
7.74***
Adjusted R-Square
0.8395
F-Value
37.96
***, **, * indicate estimates are significant at 1%, 5% and 10% per cent,
respectively.
25
Household Assets
Households with tangible assets can use those assets to improve their welfare,
both by using the asset to help the household work more efficiently and increase
their income, or through the ability to sell off the assets when the household
experiences a shock or there is a downturn in the economy (Ganesh, 2006).
Moreover, the amount of assets owned by a household reflects the income
potential of the household. In this study, too, households that have large asset
profile are assumed to be less vulnerable, because they have more production
and consumption options that increase their welfare. In the analysis, land size,
livestock and non-land household assets are included as regressors. However,
the result of the study shows that the estimated value of assets owned by the
household, other than land and livestock, is found to be an insignificant
determinant of vulnerability.
On the other hand, the amount of land size and livestock owned by households
is found to be an important determinant of vulnerability to poverty. According to
Yirga (2007), cited in Deressa et al., 2008a, livestock holding plays an important
role by serving as a store of values, and thus an insurance against risks. A
similar pattern emerges from this study, in that the amount of livestock reared by
households has a negative impact on household vulnerability level. The analysis
indicates that as livestock holding increases by one unit, household vulnerability
to poverty is reduced by 0.4 per cent and the estimate is significant at 1 per cent
significance level. The result confirms the argument that livestock holding is a
hedge against risks, because households with higher endowment of livestock
are less vulnerable and their consumption volatility is less.
Another important asset held by households is farmland on which they conduct
their production. Households in rural areas heavily depend on land for their
livelihood. It directly affects the poverty status of households since it indicates
their income potential. The result of this study also confirms this by revealing that
households that have larger land size are less vulnerable to poverty. As land
size increases by one hectare, vulnerability of households decreases by 4.1 per
cent. More importantly, the size of irrigable land owned by the household is also
found to be negatively related with household vulnerability. According to the
estimation, as irrigable land size increases by one hectare household
vulnerability level decreases by 27.65 per cent and the result is significant at 1
per cent significance level. This reinforces the fact that irrigation is related with
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
high consumption and it is an important tool for both poverty and vulnerability
reduction.
Non-farm Income and Input Use
Involvement in non-farm income generating activities is also another source of
income for the rural poor and reduces the incidence of vulnerability. Sale of
charcoal, land rental income, petty trade and manual labour are important
sources of income in the study areas. The involvement of households in nonfarm income activities reduce household susceptibility to fall below poverty line.
In the model, non-farm income is found to have an influential impact on the level
of vulnerability. Households that have alternative source of income other than
agriculture show less exposure to poverty. The analysis shows households that
have a reasonable non farm income are 15.47 per cent less vulnerable than
households with small or no non-farm income.
Use of agricultural inputs such as fertilizer and extension visits are other
important variables that define household exposure to poverty. Using fertilizer
increases productivity and thus reduces downward fluctuation of production,
which, in turn, reduces vulnerability to poverty. Extension visits, on the other
hand, increase productivity of farmers by increasing their exposure to labour and
land augmenting technologies. According to our estimation, use of fertilizer and
extension visits reduce household vulnerability to poverty by 0.06 and 1.5 per
cent respectively.
Access to Services and Capital
Proximity to infrastructures is an important factor that affects household welfare.
All-weather roads, schooling, market, input shops etc. are among major social
infrastructures that have strong linkage with poverty and vulnerability. Indicators
that have been used for access to markets, roads and services in different
studies include the distance or walking time to the nearest woreda town, market,
all-weather roads, input supply shops etc. (Chamberlin et al., 2006). In this
study, market access is measured as whether a household travels less than 16
km to and from market centres or more. The result shows that households that
are closer to market centres are less vulnerable to poverty. The analysis shows
that households that travel less than 8km (less than16 km for two ways distance)
are 1.46 per cent less vulnerable than households that travel more than 8km.
27
Proximity to all-weather roads has also similar impact on household vulnerability.
Households that travel less than 4 km to get all-weather roads are 1.77 per cent
less vulnerable than households that travel more than 4km. However, proximity
to input shops is found to be an insignificant determinant of household
vulnerability.
Social institutions such as idder, ekkub7, and number of relatives are other
variables we have to look into as far as vulnerability is concerned. Strong social
capital guarantees better access to resources and serves as a hedge against
risks. Theoretically, we expect households that have strong social capital are
less vulnerable to poverty. The result of the study also agrees with this
argument. Vulnerability estimate of households that are members of iddir and
ekkub are 34.9 and 52.34 per cents lower than households who are not
members. This is not a coincidence as membership to these institutions is
indicative of the presence of a household social capital that enhances household
resource potential which is useful to avoid vulnerability.
Rainfall, Temperature and Soil
Variation in rainfall, temperature and soil is hypothesized to significantly explain
household level of vulnerability. Since most of the agriculture in the study area is
rain fed, shortage of rain during a production season reduces production and as
a result increases households’ vulnerability to poverty. Similarly, increase in
mean maximum and minimum temperatures increases respiration loss of plants,
which, in turn, increases water requirements of plants. Therefore, increase in
temperature increases vulnerability of farmers to poverty line by reducing crop
productivity, especially during rainfall shortage. Having this in mind, change of
production season rainfall from long run average (production season average
rainfall less long run production season average) was included in the model. The
result shows that increase in production season average rainfall above long run
average reduces household vulnerability to poverty. As it can be seen from
Table 3, one millimetre increase in season rainfall above long run average
reduces household vulnerability by approximately 25 per cent. On the other
hand, increase in season mean maximum and minimum temperature above long
7
Iddir is community based institution established for mutual support and ceremonial activities
(including burial), where members are expected to raise money and help each other during
emergencies. Ekkub is traditional saving scheme in which members raise money per interval of
time (e.g. Week, month) and give for a member by turn or chance in each interval of time.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
run average increases household vulnerability to poverty. For example, 1 0C
increase in minimum temperature above long run average increases household
vulnerability to poverty by 47.20 per cent8. The result indicates that relatively
hotter areas are more vulnerable than areas with relatively cooler temperature.
Occurrence of drought is another climatic factor that affects household
vulnerability to poverty. Key informant interview has shown that there was at
least one drought occurrence during the last 20 years in the survey areas.
Drought in this context is a situation where there is severe shortage of rainfall
that resulted in complete loss of seasonal output. The analysis shows that
households that experience more frequency of droughts are found to be highly
vulnerable to poverty. As frequency of drought increases by one, household
vulnerability to poverty also increases by 31.23 per cent and the estimate is
statistically significant at 1 per cent significance level.
There are certainly other factors affecting agricultural potential at the local level
besides rainfall and temperature. Soil characteristics, in particular, are likely to
be important at the community and farm levels (Chamberlin et al., 2006). The
spatial variation in these characteristics is a major environmental factor that
accelerates or reduces household vulnerability to poverty. Soils differ in their
texture and capacity to retain moisture that is valuable for crop production. In the
analysis, three major types of soil were included; which are sandy loam, Vertisos
and Andosols. These soil dummies show strong relationship with household
extent of vulnerability. For example, households which are in areas of Vertisols
are more vulnerable than households that are found in areas of sandy loam soil.
This is because sandy loam soil has relatively better water retaining capacity
than Vertisols and it can give yields with less moisture with the help of dew once
the plants are grown to a certain degree. But, in the case of Vertisols, it has the
property of cracking in time of rainfall shortage and it becomes the major reason
for crop failure. The results of the study show that households in areas of
Vertisols are 25.47 per cent more vulnerable to poverty than areas with sandy
loam soil.
8
Mean maximum temperature and Andosols were excluded due to higher collineraity with minimum
temperature and soil dummies.
29
Non climatic Shocks
Shocks that cause income or asset losses are also likely to reduce consumption
if credit constraints are binding or if the shock reduces expected life-time
earnings by destroying the household’s asset base (Tesliuc and Lindert, 2002).
Household experiences of shocks have, thus, more things to do with household
vulnerability to poverty. In the survey, households reported different types of
non-climatic shocks that include input price rise, death and illness of a
household member, death of animal, crop pest and disease and output price fall
etc. Experiencing one or more of these shocks results in reduction of household
welfare, because they negatively affect household production potential.
However, in this study, the distribution of shocks between the vulnerable and
non-vulnerable households is random for most of the shocks. But there is
difference in vulnerability for households that encountered unexpected input
price rise, death of head of the household, death of animal and crop pest and
disease. For instance, 71 of the vulnerable households have reported input price
rise whereas only 15 per cent of the non-vulnerable households have
encountered input price rise, and the difference is significant at 1 per cent
significance level.
According to the econometric analysis, death of a household head and input
price rise negatively affect household welfare. Households that face death of a
head and input price rise are 75.25 and 9.47 per cent more vulnerable than
households that do not face any of these shocks. Other shocks such as output
price fall, death of animal, illness of a household member and incidence of crop
pest and disease are found to be statistically significant.
5. CONCLUSION AND POLICY IMPLICATIONS
Based on the household survey data in Adama and Lume Woredas, this study
analyzed the probability of farmers falling below consumption poverty line due to
climatic conditions and shocks namely mean seasonal rainfall, temperature and
frequency of drought. In this analysis, household consumption expenditure that
consists of both food and non-food expenditure are used as proxy for welfare.
Like poverty analysis, vulnerability of households is attributed to both household
and environmental characteristics. By assuming consumption is log normally
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
distributed, a household probability of falling below poverty line of USD1 is
estimated using the Vulnerability as Expected Poverty approach.
Estimates show that about 62 per cent of the households are observed to be
poor during the survey. Computed using 0.5 as a threshold above which a
household is called vulnerable, about 68 per cent of the households are also
vulnerable to poverty during the coming year. An attempt to show the sources of
vulnerability indicates that about 52 per cent of the households are vulnerable to
poverty due to low consumption mean and about 16 per cent of them are
vulnerable due to high consumption volatility.
Moreover, the study has attempted to track the correlates of household
vulnerability to poverty by assuming household vulnerability is linearly related
with household and environmental characteristics. It was observed that a
household head’s age and education level, land size, livestock size, proximity to
roads and market reduce household vulnerability to climate change, whereas
family size and experiencing shocks tend to increase household vulnerability to
climate change. Use of inputs such as fertilizer and extension services, access
to irrigation and non-farm income also reduces household vulnerability to
poverty. Of particular importance to this study is that all climate and environment
related factors are found to affect household vulnerability to poverty. For
instance, increase in mean seasonal rainfall above long run average reduces
household vulnerability to poverty, whereas increase in mean minimum
temperature above long run average increases household exposure to poverty.
Moreover, the nature of soil is also related with vulnerability to poverty. Sandy
loam soil was found to reduce household vulnerability to poverty but Vertisols
tend to aggravate household vulnerability to poverty. On the other hand, use of
one or more adaptation method was found to reduce the incidence of
vulnerability of households.
The most important lesson from this study is that increasing mean consumption
(income) of households is not the only way to reduce vulnerability, but reducing
volatility of consumption is also important. This means enabling farmers to meet
the daily minimum requirement is not enough by itself, unless there are
interventions that reduce volatility of consumption. This study proposes that
expansion of irrigation use, off-farm income opportunities, use of fertilizer and
extension services are possible intervention areas to reduce household
vulnerability.
31
In addition to policy interventions to increase consumption and reduce volatility
of consumption, promotion of adoption of different adaptation methods such as
use of early maturing crops, soil conservation techniques and planting trees can
reduce vulnerability. In case of occurrence of drought, coping mechanisms such
as keeping livestock and creation of non-farm income opportunities reduce
vulnerability.
Finally, strengthening the ability of households to reduce, mitigate or cope with
the effects of both climatic and non-climatic shocks is likely to reduce their
vulnerability to poverty. However, these results should be interpreted and used,
with the recognition of possible errors in consumption measurement and
statistical problems that may arise with spatial variables.
Analysis of Household Vulnerability to Climate Change and Adaptation Options:…
References
Adane, N. and E. Bezabih. 2003, Rural Poverty In Central Highland Of Ethiopia:The Case
Of Bereh-Aleltu District. AESE 7th Annual Conference Proceedings, Addis Ababa,
Ethiopia.
Adger, W. and N. Kelly (1999) “Social Vulnerability to Climate Change and the
Architecture of Entitlements,” Global Climate change, 253-266.
Antle, J S. Capalbo, E. Elliot and K. Paustian (2004) “Adaptation, Spatial Heterogeneity,
and the Vulnerability of Agricultural Systems to Climate Change and CO 2
Fertilization: An Integrated Assessment Approach,” Climate change, 64: 289-315.
Asiimwe, J. B. and P. Mpuga (2007), Implication of Rainfall Shocks for Houseehold
Income and Consumption in Uganda. AERC Research Paper No. 168. African
Economic Resarch Consortium, Nairobi, Kenya.
Befekadu D. and N. Berhanu (2000) Annual Report on the Ethiopian Economy. Ethiopia
Economic Association, Addis Ababa, Ethiopia.
Chamberlin, J, J. Pender and B. Yu (2006), Development Domainsfor Ethioipa: Capturing
The Geographical Context Of Smallholder Developing Options. Paper submitted for
ESSP Policy confrence, Addis Ababa, Ethiopia, 6-8 june, 2006.
Chaudhuri, S. J. Jalan and A. Suryahadi (2002), Assessing Household Vulnerability to
Poverty from Cross-sectional Data: A Methodology and Estimates from Indonesia.
Columbia University, Department of Economics Discussion Paper 102-52. New York.
Chaudhuri, S. (2003), Assessing Vulnerability to Poverty: Concepts, Emprical Methods
and Illustrative Examples. Columbia University, Department of Economics
(unpublished)
Cline, W. (1996) "The Impact of Climate Change on Agriculture: Comment," American
Economic Review, 1996: 1309-11.
Dercon S. and J. Hoddinott (2008), Collective Action and Vulnerability: Burial Socities in
Ethiopia CAPRi Working Paper No.83 Washington, DC.
Deressa, T. (2007), Measuring The Economic Impact of Climate Change on Ethiopian
Agriculture: Ricardian Approach. World Bank Policy Research Working Paper No.
4342. Washington, DC.
Deressa T, R. M. Hassen, T. Alemu, M. Yesuf and C. Ringler (2008), Analyzing the
Determinants Of Farmers’ Choice of Adaptation Methods and Perception on Climate
Change in the Nile Basin of Ethiopia. IFPRI Discussion Papper 00798. IFPRI,
Washington, DC.
Deressa T and R. M. Hassan and C. Ringler. 2008. Measuring Ethiopian Farmers’
Vulnerability to Climate Change Across Regional States. IFPRI Discussion Papper
No. 00806. IFPRI, Washington, DC.
Deressa, T, R. M. Hassan, and R. Claudia (2008), Assessing Household Vulnerability to
Climate Change: The Case of Farmers in the Nile Basin of Ethiopia. Paper presented
at IFPRI Workshop on Climate and Agriculture, Adama, Ethiopia, December 2008.
Di Falco, S. G. Kohlin and M.Yesuf (2008), Adaptation to Spatial Climatic Variability and
the Implications for Productivity and Risk Management in the Nile Basin, Ethiopia.
(publisher missed)
EEA (Ethiopian Economic Association) (2008) ‘Climate change and Development
Adaptation Measures in Ethiopia,’ Economic Focus, vol. 11 No.1.
Ganesh C. Mallick (2006), Understanding Poverty and Vulnerability: A Study among three
Indigenous Communities of Eastern India. (Unpublished report)
Gaihai, R. and K. Imai (2008), Measuring Vulnerability and Poverty: Estimates for Rural
India. UNU-WIDER Rsearch Paper No. 2008/40. UNU-WIDER.
33
Gbetibouo, G. A. and C. Ringler (2009), Mapping South African Farming Sector
Vulnerability to Climate Change and Variability: A substantial Assessment. IFPRI
discussion Paper No.00885. IFPRI, Washington, D.C.
Hassan, R and C. Nemachena (2007), Micro-Level Analysis of Farmers’ Adaptation to
Climate Change in Southern Africa. IFPRI Discussion Paper No.00714. IFPRI,
washington, DC.
IFPRI (2009), Agriculture and Climate Change : An Agenda for Negotiation in
Copenhagen. International Food Policy Research Institute 2020 Vision Focus 16.
Washington, D.C.
IFPRI (2009), Climate Change and Agriculture: Impact On Agriculture and Costs Of
Adaptation. International Food Policy Research Institute (IFPRI) Food Policy Report,
Washington, D.C.
ILRI (International Livestock Research Institute) (2006), Mapping Climate Vulnerability
and Poverty in Africa. Report to the Department for International Development. ILRI,
Nairobi, Kenya.
IPCC (2001), Overview of Impacts, Adaptation, and Vulnerability to Climate Change. At
http/www.ippc.ch/
Jamal, H. (2009), Assessing Vulnerability to Poverty: Evidence from Pakistan. Social
Policy and Development Centre (SPDC) Research Report No.80.
Ligon, E and L. Schechter. 2003. “Measuring Vulnerability”. The Economic Journal , C95C102.
MoFED (Ministry of Finanace and Economic Development) (2008), Dynamics of Growth
and Poverty in Ethiopia. Addis Ababa, Ethiopia.
NMSA (National Meteorological Service Agency) (2001), Initial National Communication
of Ethiopia to the United Nations Framework Convention of climate Change. MoWR,
Addis Ababa, Ethiopia.
Olmos, S. (2001), Vulnerability and Adaptation to Climate change: Concepts, Issues,
Assessment Methods. Climate Change Knowledge Network foundation Paper, Oslo.
Padgham, J. (2009), Agricultural Development Under A Changing Climate:Opportunities
and Challenges for Adaptation. World Bank, Washington, DC
Quisumbing, A and J. Hodinnott (2003), Methods for Microeconometric Risk and
Vulnerability Assessments. World Bank Social Protection Discussion Paper Series
0324. World Bank, Washington, DC.
Raghav, G. and I. Katsushi (2008), Measuring Vulnerability and Poverty:Estimates for
Rural India. UNU-WIDER Research Paper No. 2008/40. United Nations UniversityWIDER.
Shewmake, S. (2008), Vulnerability and the Impact of Climate Change in South Africa
Limpopo River Basin. IFPRI Discussion Paper No. 00804. IFPRI, Wasington, DC.
Tesliuc, E. and K. Lindert (2002) Vulnerability: A Quantitative and Qualitative
Assessment. Guatamela Poverty Assessment Program No.36209, Guataela.
World Bank (2008), Ethiopia: Climate Risk Fact Sheet. World Bank, Washington, DC.
Yesuf, M. S. Di Falco, T. Deressa, C. Ringler and G. Kohlin (2008), The Impact of Climate
Change and Adaptation on Food Production in Low-Income Countries Evidence from
the Nile Basin, Ethiopia. IFPRI Discussion Paper No. 00828. IFPRI, Washington, DC.